Now accepting guest editorial from SAP, Oracle, Workday and ServiceNow practitioners.Pitch a guide
The Sign-Off Table

Lesson 4 of 5

Waves and cutover

9 min

Listen to this lesson

0:00 / 6:44

A PO migration is not one cutover weekend. It is a series of small switches, one interface or one group at a time, while PO and Integration Suite run side by side. Waves group those switches so that the team learns early, partners are coordinated once, and the business never has too many changes at the same time.

Designing waves

  • Pilot wave first: three to five interfaces that are low criticality, Ready to migrate, and cover the main adapter types. The goal is to prove the tooling, transports, monitoring and test approach, not to move volume.
  • Group by partner where certificates, IP allowlists or endpoints change, so each partner is contacted and tested once.
  • Group by business process where process owners sign off, so Dan (finance) or Maria (O2C key user) sign once per wave, not per interface.
  • Put critical, high-volume interfaces in the middle, after the pilot has proven the method and before deadline pressure builds.
  • Leave redesigns with external dependencies to the end, with buffer before the maintenance date.
  • Respect release calendars: avoid S/4HANA release weekends, period close and peak shipping seasons.
  1. Go-live + 3 monthsPilot wave4 interfaces: two internal IDoc flows, one SFTP supplier, one SOAP service. Method proven.
  2. Go-live + 4 to 5 monthsWave 1: procure-to-pay14 interfaces, grouped by supplier; Lakeside switched in one coordinated change.
  3. Go-live + 6 to 7 monthsWave 2: order-to-cash and logistics18 interfaces, including the consolidated carrier flow.
  4. Go-live + 8 to 9 monthsWave 3: finance and bank11 interfaces, avoiding quarter-end close; bank certificates exchanged in advance.
  5. Go-live + 10 to 12 monthsWave 4: redesigns and retirementsRemaining redesigns; 7 retirements confirmed after year-end; PO decommission planned.
Contoso's migration waves, finishing by go-live + 12 months, well ahead of the end of 2027.
The Sign-Off Table · thesignofftable.com/academy/sap-int-301

Coexistence and parallel run

During coexistence, each interface runs on exactly one platform at a time in production. A parallel run here means running the new flow alongside the old one on the same real inputs without the new flow delivering to the real receiver: its output goes to a comparison target. That gives the strongest equivalence evidence (lesson 5) without duplicate business documents. A true dual delivery, where both platforms send to the partner, almost always creates duplicates and should be avoided.

One interface during coexistence: it delivers from exactly one platform at any moment, and the shadow run never delivers to the partner.
The Sign-Off Table · thesignofftable.com/academy/sap-int-301
  1. Production inputCMP sends the ORDERS IDoc as today.
  2. PO deliversXP1 maps and delivers to Lakeside as the system of record.
  3. Copy to new flowA copy of the same input goes to the new Integration Suite flow.
  4. New flow output capturedOutput written to a comparison store, not to Lakeside.
  5. CompareOutputs compared daily; differences explained or fixed before the switch.
A shadow parallel run that proves equivalence without duplicate orders.
The Sign-Off Table · thesignofftable.com/academy/sap-int-301

Cutover per interface

Each interface switch is a small cutover with its own runbook. For an outbound interface from S/4HANA: deploy and test the flow in production, stop the PO channel, repoint the sender (for example the RFC destination or port in CMP) to Integration Suite, confirm the partner's allowlists and certificates, send the first real message and check it end to end. The rollback is the reverse repointing, and it must be tested too. Keep the PO configuration intact until the wave's hypercare ends.

The runbook for switching one outbound interface, with a rollback that only works because the PO configuration stays intact.
The Sign-Off Table · thesignofftable.com/academy/sap-int-301

What would you do?

Wave 1 is set to switch Lakeside's order interface on Saturday. On Thursday, Lakeside says its firewall team needs ten working days to allow Integration Suite's IP addresses. What should Luis, as cutover manager, decide?

Key points
  • Start with a pilot wave that proves the method; group later waves by partner and process.
  • In production each interface runs on one platform; use shadow parallel runs, not dual delivery.
  • Every interface switch has a runbook with a tested rollback by repointing.
  • Engage partners at wave planning; their lead times set the schedule.

Check your understanding · not graded

Lesson check

  1. 1. What is the main purpose of a pilot wave?

  2. 2. Why avoid both platforms delivering to the same partner during a parallel run?

  3. 3. What is the usual rollback for a switched outbound interface?

  4. 4. Why group interfaces by partner within waves?